Announcing the Agentic Catalog Experience in Amazon Quick

來源原標題

Announcing the Agentic Catalog Experience in Amazon Quick

作者

Srikanth Baheti

原始標籤

Amazon Quick Suite · Announcements · Foundational (100)

原始連結

https://aws.amazon.com/blogs/machine-learning/announcing-the-agentic-catalog-experience-in-amazon-quick/

摘要原文

As organizations embrace AI-powered analytics, the value of a natural language (Text2SQL) answer is only as good as the business context behind it. We’re entering a phase where semantic richness (table and column descriptions, and relationships) must flow directly from where it’s authored in upstream data catalogs and semantic tools into the AI products that serve end users. Products like Amazon Quick can no longer operate in isolation. They need to natively consume and reason over the definitions, relationships, and governance metadata that data teams curate in systems like AWS Glue Data Catalog and Databricks Unity Catalog. This shift from siloed metadata to connected, catalog-aware AI is what enables intelligent analytics at scale. Enterprise data teams have done the hard work. They have invested heavily in upstream catalog platforms such as AWS Glue, Databricks Unity Catalog, Snowflake Horizon, Collibra, and dbt. On these platforms, they meticulously define table descriptions, column semantics, primary and foreign key relationships, glossary terms, and metric definitions. Yet when it comes to enabling end users (such as sales managers, marketing directors, and finance leads) for production-ready AI and trusted dashboards, a significant gap remains. When data curators (business intelligence engineers, analytics leads, and senior analysts) need to enable their business users in Amazon Quick, they face three compounding challenges: The problem isn’t upstream. The metadata exists. The governance is defined. The relationships are mapped. The problem is the last mile: translating that rich catalog context into a curated, consumable experience that delivers grounded AI answers and deterministic dashboards end users can trust. Today, we’re announcing the Agentic Catalog Experience in Amazon Quick, an AI-powered workflow that helps data curators rapidly define their context boundary, inherit upstream semantics, and enable end users for grounded Q&A and trusted dashboards at scale. At the heart of this experience is the Quick Agent, scoped to discovery, creation, and inheritance tasks within the catalog context. It uses the semantic context from the catalog connection to summarize the entire catalog at a glance, engage the customer in natural language conversation, surface the most relevant tables and relationships based on the customer’s use case, and assess metadata readiness. Then, with a single conversational confirmation, it auto-creates Catalog-Generated Datasets and Topics with targeted metadata inherited from the upstream catalog. No manual configuration. No context-switching. No weeks of setup. Instead of scrolling through thousands of tables to find the right ones, curators use natural language. With the Agentic Catalog Experience, curators describe what they need: Curator: “I’m a Senior Analyst on the Finance team. I need tables for quarterly revenue reporting and cost analysis.” The Quick Agent searches across your entire catalog to surface the most relevant tables instantly, using all available metadata including business descriptions, tags, Gold/Silver/Bronze classifications, quality scores, table health scores, and glossary terms. No more manual browsing. No more guessing. After the curator selects their tables, the Quick Agent creates catalog representations (Datasets) at scale in a single guided workflow. Your upstream catalog remains the source of truth because the default creation path is Direct Query. Datasets with inherited semantics are flagged with a clear “Semantics Inherited” badge, and their metadata is read-only. Authors can refresh inherited metadata on demand by choosing the sync button to stay aligned with their catalog. Quick Agent: “Creating 6 Catalog-Generated Datasets now: revenue_by_region created (DirectQuery, read-only metadata), cost_centers created, and gl_transactions created.” The Quick Agent carries forward targeted metadata from your catalog into the assets it creates. Today, inheritance is deliberately focused on two key areas to avoid noise and keep Datasets clean: Note : While all available metadata (Gold/Silver classifications, quality scores, tags, and health scores) is used during discovery to find the right tables, inheritance into Datasets is intentionally scoped to table and column definitions today. We plan to add more metadata types to Datasets over time. Quick Agent: “I detected 3 relationships between these tables and created a Topic called ‘Finance Revenue Model’ with the star schema joins preconfigured. Table and column definitions have been inherited from the upstream catalog.” The curated Datasets and Topics are ready for use immediately: After creation, the metadata tied to these Datasets and Topics feeds into the Amazon Quick semantic store, which powers re-ranking and unified context for AI-powered Q&A. Getting from catalog connection to the first business question takes minutes, not weeks. A key design principle underpins this experience: Amazon Quick is a consumer of upstream catalog metadata, not a dedicated catalog itself. This means: Support for additional catalog platforms is coming soon. Metadata inheritance is intentionally focused to keep Datasets clean and production-ready: Here’s what this means for the business users downstream: A sales manager asks: “What were our Q4 sales by region?” Behind the scenes, the AI agent: No manual dataset configuration required. The curator defined the context boundary once with the Quick Agent, and every end user benefits immediately. The Agentic Catalog Experience doesn’t exist in isolation. Combined with the broader platform capabilities of Amazon Quick (including integration with Slack, Outlook, documents, and knowledge bases), end users get the full enterprise context: This unified context enables production-ready AI answers, grounded in your organization’s specific data and semantics. To get started with the Agentic Catalog Experience, create a data source connection to your AWS Glue Data Catalog in Amazon Quick. After you establish the connection, the Quick Agent guides you through discovery, schema exploration, and Topic creation in a single conversational workflow. In this walkthrough, we connect to a Glue Data Catalog and build a Financial Analytics Topic. In Amazon Quick, create a new data source. From the list of connection types, select Glue Data Catalog (available in preview), and then choose Next . This connection is for the metadata. With it, Amazon Quick can consume the table and column definitions and the relationships your teams have already curated in AWS Glue. Figure 1: Selecting the Glue Data Catalog connection type in Amazon Quick A Glue Data Catalog connection works together with an Amazon Athena connection. Glue provides the metadata, and Athena provides the query path to the data itself in Amazon Simple Storage Service (Amazon S3). Create the Athena data source as well, so that Amazon Quick can run queries against the underlying data. After you create both, the Data sources page shows the two entries side by side: the Glue Data Catalog source for the metadata and the Athena source for the data. Figure 2: The Glue Data Catalog and Athena data sources listed together Open the GDC-Demo data source detail page. Under Data connections , you can see the linked Athena data source that Amazon Quick uses to query the data. Choose Explore data to launch the Quick Agent scoped to this data source. Figure 3: Launching the Quick Agent from the data source detail page The Quick Agent panel opens on the right side of the screen, automatically scoped to the Glue Data Catalog data source. The “Specific data” mode is selected, with “GDC-Demo” pinned as the context boundary. As a result, the Agent surfaces only metadata from this specific catalog connection. Figure 4: The Quick Agent scoped to a specific catalog connection Ask the Agent to explore your catalog.